Re: [Quantlib-dev] Implementing ComplexArray class.

[Joseph W. <jo...@gn...>]

=E5=9C=A8 Monday 02 October 2006 03:31=EF=BC=8CLuigi Ballabio =E5=86=99=E9=
=81=93=EF=BC=9A
> On Sun, 2006-10-01 at 18:44 -0500, Joseph Wang wrote:
> > Looking over Numerical Recipes in C++, they suggest implementing a list
> > of N complex numbers as an array of 2*N floats because they argue that
> > you maintain control of how the complex number calculations are
> > implemented.
>
> That pretty much defeats the purpose of having a std::complex class in
> the standard library, doesn't it?  It's like arguing that one should
> implement his own containers to maintain control.

That was my first reaction.

Any objections to creating an implementation agnostic ComplexArray class th=
at=20
using valarray and complex?  The reason for creating a separate class is th=
at=20
I'd like to have a nice interface that interoperates with the Array class.